Stephanie Bauer doesnât fit the spammer profile. Sheâs not sending e-mails about get-rich-quick schemes or Nigerian bank accounts. And she doesnât mail anyone without confirming that the address is double opt-in.
Yet Bauer, who manages three e-mail newsletters for the Cellular Telecommunications & Internet Association, Washington, D.C., said she has been treated as a spammer more than once because spam filters have blocked her e-mail newsletters.
Responding to complaints about the deluge of unwanted commercial e-mail, ISPs and individual companies have installed software products that detect suspicious or fraudulent e-mail messages.
Many of these filters use content-based rules that search for common words or phrases used in spam, such as "free" and "sex," as well as capital letters, underlining and excessive use of various font sizes and colors. Filters can also be volume-sensitive, built to respond to high volumes of messages, which are a common sign of bulk, unsolicited e-mails; they bounce e-mails back to the sender labeled "too many recipients."
Thereâs little question that spam has reached epidemic proportions. The Radicati Group Inc., a market research consultancy in Palo Alto, Calif., said 32% of all e-mail messages in 2002 were spam, and by the end of 2003, the percentage will increase to 45%. The cost of spam for a company with 100 employees who receive an average of 10 spam e-mails per day is $7,986 per year, according to Computer Mail Services Inc.âs Spam Calculator.
Dreaded false positives
Unfortunately for marketers, the filters can be indiscriminant, mistakenly blocking legitimate messages. The phenomenon is known as a false positive.
Some experts believe ISPs incorrectly treat one of six permission-based e-mails as spam and that 15% of messages do not get through to customersâ in-boxes.
In fact, the problem may be worse for b-to-b marketers. Thatâs because, unlike the majority of consumer e-mail (which is handled by a handful of ISPs), the b-to-b space is fragmented, with many more ISPs.
Spam filter rules, configured at the server level, can be dictated by the ISP or software systems deployed behind a corporate firewall. Especially problematic are filters set up by individual companies by overworked IT departments.
In Bauerâs case, false positives have meant lost customers. "Itâs destroying our bottom line," she said.
CTIAâs primary paid news product, CTIA Select News & Information Service, was launched in January 2002 and commands a $269 yearly subscription. But Bauer estimated the association has lost, to date, 15 to 20 of its 600 paying subscribers due to faulty spam filters.
In the last six months alone, subscriptions have dipped by about 5,000âa number that includes free and paid e-mail products. (In fairness, Bauer attributes half of the falloff to filters and half to the wireless industryâs massive layoffs.)
Bauer and her e-mail provider, Denver-based Experian Corp., have been working feverishly with ISPs like AT&T, which tend to cooperate when problems have been discovered, and with client companies whose servers inadvertently block messages. But some subscribers have simply walked away.
"We have people who donât want to bother because itâs too much work to renew, so they just unsubscribe," she said.
Meanwhile, handling nondelivery on a one-to-one basis is unwieldy, impractical and cost-prohibitive. "With a list of 35,000 to 40,000, if 10% arenât getting it, you canât work with each individual problem," Bauer said.
Filters that check content in subject line, the "from" address, "reply to" address and the message body are prone to make mistakes because the same key words that have been flagged as often appearing in spam may be the same words contained in a marketerâs message.
Josh Carlson, Web manager at The Insurance Journal, sends four free, opt-in newsletters to insurance professionals. He said content often contains words such as "insurance," "fraud" and "illness"âwords that anti-spam filters mistakenly identify as red flags.
Brightmail Inc., an anti-spam software provider based in San Francisco, said it filters 50 billion messages per month for 600 customers. Enrique Salem, president-CEO of Brightmail, said content-based filtering systems "are flawed. Itâs the wrong way to filter spam."
Brightmail uses a "probe network," wherein decoy e-mail accounts are set up across the Internet, and Brightmail technicians aggregate messages in the probes. Brightmail sends that information to software running on customersâ sites. "Our false positive rate is one in a million," Salem said.
But there are no easy answers.
"I donât think thereâs anyone in the industry that doesnât have this problem," said Dave Lewis, VP-deliverability management and ISP relations at San Mateo, Calif.-based Digital Impact Inc. "I donât care how conservative a mailer is."
Lewis and other experts agree the problems have sharply increased in the past 60 to 90 days as major ISPsâincluding AOL, Microsoft Corp.âs MSN and Yahoo!âhave unveiled new anti-spam tools.
Most e-mail providers, including companies like Bigfoot Interactive, Digital Impact and ExactTarget, review clientsâ e-mail creative against attributes commonly found in spam.
"The goal is to not have too many of those attributes," said Michael Della Penna, chief marketing officer at New York-based Bigfoot Interactive and co-chair of the Association for Interactive Marketingâs Council for Responsible E-Mail. The council consults with clients through its "delivery lab" to address those issues.
The Network Advertising Initiative formed an E-mail Service Provider Coalition in January to respond to deliverability problems experienced by legitimate marketers.
The group includes 28 stakeholders, including Acxiom Corp., Digital Impact Inc. and DoubleClick Inc. In February, the coalition launched a Yahoo! discussion group called "I Did Not Get My E-mail" to provide a forum for consumers and businesses alike.